I recently purchased an Advantest R6142, and, having played with it a bit, would like to share what I’ve found.
First of all, the R6142 is, despite appearances, a complete redesign of the older Takada/Advantest TR6142. The board layouts have completely changed, the TR6142 twiddle pots have been replaced with software calibration, and even the reference voltage has been changed. This isn’t to say that the TR6142 documentation is useless. It provided complete schematics and a very detailed theory of operation, both of which the newer R6142 documentation omit. I suspect that there are enough similarities that older “TR” documentation can provide insight into how the newer R6142 works.
(BTW, one lonely pot DOES remain in the R6142 – I think it’s the equivalent of the TR’s R80, which provides linearity correction “by correcting the time constant errors caused by the turn-on and turn-off resistances” of the transistors used to implement the voltage-generating duty cycle. I haven’t had the guts to play with this, since I really don’t know how to measure its effects, and the manual doesn’t provide a procedure.)
I heard it is possible to hack R6142 into a R6144 by shorting JP1, JP2 and JP11.
Thanks for the tip, baoshi! It is indeed true – all you need to do is wave your soldering iron over the R6142, and it will magically turn into an R6144!! (Well, the front panel labels won’t change, but functionally, it literally BECOMES a 6144, and even identifies itself as such while powering on!)
Initially curious about this ‘upgrade’, I noticed that there were two characteristics of the two models that lent credence to the rumor – 1) the parts lists for both units were (so far as I could tell) absolutely identical, and 2) the specifications (accuracy, tempco, etc) for the two units are absolutely identical. The only difference is that the R6144 extends the voltage range from 16 to 32 volts, and extends the current range from 120ma to 160ma. These differences seem hardly sufficient to justify a completely different model and design, especially given the identical performance.
The R6142 comes with U shaped jumpers installed for JP1, JP2, and JP11. But they are cut! So I’m guessing that all of the units are built and originally tested as R6144’s, but then downgraded to R6142’s by cutting the jumpers. (If I remember correctly, one of them was on the digital board – probably something that the CPU can read to determine personality, and the others were near voltage sources – possibly to raise the upper limit of operating voltages/currents.) Here are the jumper locations (in Japanese).
http://bbs.38hot.net/thread-15339-1-1.htmlIn playing with the R6142 (prior to its promotion to 6144), I came across a peculiarity that’s worth knowing about.
The front panel allows simple manual stepping by a fixed amount (1mv in this case) by pressing the appropriate up-arrow key. It also allows for an automated sweep, simply by pressing and holding the same key. While trying this single stepping, the unit generally behaved as expected – press the button once, and the output increased by 1mv. But every once in a while, it curiously jumped by 2mv. Having noticed this, I set up a sweep, and recorded the voltages output.
The charts below show the results over about the first 1000 1mv steps (0n 10 volt range). The first chart shows that, overall the output appears very linear, as expected. The second chart shows the deviation between the expected voltage and the actual output. Clearly, near 0V output, the output is low by about 3mv. But then the output deviation settles into a curious sawtooth pattern. Taking a look at the magnitude of each step (third chart) shows what’s going on. Each voltage step is about 1mv, as expected. But periodically (about every 126 steps) the step size jumps up to 2mv – just for a single step.
This ‘occasional double step’ behavior seems to occur on all voltage ranges. It is presumably due to some occasional ‘catch up’ being done in setting the digital duty-cycle that generates the reference voltage. I didn’t try it in current mode, but suspect it would behave similarly. Having uncovered this behavior, I just thought I'd share it...